Abstract
Decades of research into the etiology of conduct disorder (CD) has yet to yield a consensus on the existence of sex differences in underlying genetic and environmental influences. This may be partly due to the failure of many previous studies to make a distinction between non-aggressive and aggressive CD symptoms or test for potential developmental changes in sex differences in the etiology of conduct problems. To address these gaps, we fit a series of univariate and bivariate biometric sex-difference models to self-reported non-aggressive and aggressive CD symptoms in a community-based sample of twins (N = 1548, ages 9-17 year), grouped into ages 9-13 and 14-17 years. Relative model fit was evaluated using the Bayesian Information Criterion (BIC), which favors parsimony, and by Chi square difference tests. The univariate sex-scalar model provided the best fit to the data for both non-aggressive and aggressive CD symptoms at ages 9-13 and 14-17 years. Thus, the same genetic and environmental factors influenced CD symptoms in both sexes, but the total variability was lower in females than males. At both ages, the heritability of non-aggressive CD symptoms was lower than heritability of aggressive CD symptoms, and shared environmental effects were only observed for non-aggressive CD symptoms. However, estimates for genetic and environmental factors could be not be constrained to be equal across age groups for either CD subtype, suggesting substantive developmental changes in the relative influence of genetic and environmental factors on individual differences in CD symptoms. For both subtypes, the heritability was larger, and shared environmental effect smaller, in the older age group than the younger age group. A bivariate quantitative sex differences model provided the best fit to the data at ages 9-13 years. Covariation between non-aggressive and aggressive CD symptoms was due to overlapping shared and non-shared environmental factors in males and females but the overall covariation was greater in males than females. In contrast, at ages 14-17 years, the sex-scalar bivariate model provided the best fit to the data, and covariation between non-aggressive and aggressive CD symptoms was due to overlapping genetic and non-shared environmental factors. Thus, the etiology of self-reported conduct disorder varied substantially by symptom type and age. However, quantitative sex differences were only apparent when the covariation between the two subtypes was considered.